The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
258
Spectral variables
Characteristic of the plant related with the
variable/index
Definition
Described by
A-1D, B-1D, C-1D, D-1D, E-1D,
F-1D,G-1D, H-1D, I-1D.J-1D,
maximum 1st derivative spectra
Pigments absorption in visible region and
water, cellulose, starch and lignin absorption
inNIRand SWIR.
Maximum 1st derivative spectra of 10 'slopes' :
blue edge , yellow edge and red edge and other
7 'slopesi in NIR and SWIR regions
Gong et al., 2002
Pu et al., 2004
A-WP, B-WP, C-WP, D-WP, E-
WP, F-WP, G-WP, H-WP, I-
WP.J-WP, spectral position
variables corresponding "IDs"
Pigments absorption in visible region and
water, cellulose, starch and lignin absorption
in NIR and SWIR.
Corresponding spectral positions of "IDs" of 10
'slopes' : blue edge , yellow edge and red edge
and other 7 'slopesi in NIR and SWIR regions
Gong et al., 2002
Pu et al., 2004
R550
Chlorophyll content
Reflectance at 550 nm
Thomas and Gausman, 1977
R680
Chlorophyll content
Reflectance at 680 nm
Thomas and Gausman, 1977
Wl, Water Index
Water status
R900/R970
Periuelas et al., 1997
NDVI, Normalized Difference
Vegetation Index
Photosynthetic area; cell structure multi-
reflected spectra
(Rnir-R r )/(Rnir+Rr)
Rouse et al., 1973
SR, Simple Ratio
Same as NDVI
Rnir/Rr
Jordan, 1969
PRI, Photochemical Reflectance
Index
Water stress
( R531^570 V ( R531 + R57O )
Thenot et al., 2002
SIPI, Structural Independent
Pigment Index
Carotenoids: chlorophyll a ratio
(R 4 45-R 8 oo)/(R680‘R80o)
Periuelas and Filella, 1998
NPCI, Normalized total Pigment
to Chlorophyll Index
Senescence
( R680-R430)/(R680 + R430)
Periuelas et al., 1994
NPQI, Normalized
Phaeophytinization Index
Senescence
(R415-R435)/(R435+R4 35 )
Barnes et al., 1992
Periuelas et al., 1995
LCI, Leaf Chlorophyll Index
Chlorophyll content
(R850“R71oV(R850 + R68o)
Datt, 1999
NDWI, ND Water Index
Water status
(R860-Rl240y(R860 + Rl240)
Datt et al., 2003
Gao, 1996
DSWI, Disease water stress
Water status
(R 8 02+R547)/(Rl657 + R682)
Galvâo et al., 2005
RATI0 975 3-band ratio at 975
nm
Water status
2*R960-990/( R920-940+ Rl090-1110)
Pu et al., 2003
RATI0 120 o 3-band ratio at 1200
nm
Water status
2*Rl180-1220/( Rl090-1110+ R-1265-1285)
Pu et al., 2003
WP-975: wavelength position at
975 nm
Water absorption feature at 975 nm
See reference for the defiinition of wavelength
position at 975 nm
Pu et al., 2003
DEP-975 absorption depth at
975 nm
Water absorption feature at 975 nm
See reference for the defiinition of absorption
depth at 975 nm
Pu et al., 2003
WID-975 absorption width at
975 nm
Water absorption feature at 975 nm
See reference for the defiinition of absorption
width at 975 nm
Pu et al., 2003
AREA-975 absorption area at
975 nm
Water absorption feature at 975 nm
See reference for the defiinition of absorption
area at 975 nm
Pu et al., 2003
WP-1200: wavelength position
at 1200 nm
Water absorption feature at 1200 nm
See reference for the defiinition of wavelength
position at 1200 nm
Pu et al., 2003
DEP-1200 absorption depth at
1200 nm
Water absorption feature at 1200 nm
See reference for the defiinition of absorption
depth at 1200 nm
Pu et al., 2003
WID-1200 absorption width at
12Q0 nm
Water absorption feature at 1200 nm
See reference for the defiinition of absorption
width at 1200 nm
Pu et al., 2003
AREA-1200 absorption area at
1200 nm
Water absorption feature at 1200 nm
See reference for the defiinition of absorption
area at 1200 nm
Pu et al., 2003
WP-1750: wavelength position
at 1750 nm
Water absorption feature at 1750 nm
See reference for the defiinition of wavelength
position at 1750 nm
Tian et al., 2001
Pu et al., 2003
DEP-1750 absorption depth at
1750 nm
Water absorption feature at 1750 nm
See reference for the defiinition of absorption
depth at 1750 nm
Tian et al., 2001
Pu et al., 2003
WID-1750 absorption width at
1750 nm
Water absorption feature at 1750 nm
See reference for the defiinition of absorption
width at 1750 nm
Tian et al., 2001
Pu et al., 2003
AREA-1750 absorption area at
1750 nm
Water absorption feature at 1750 nm
See reference for the defiinition of absorption
area at 1750 nm
Tian et al., 2001
Pu et al., 2003
Table 2. Summary of 46 spectral variables extracted from the in situ hyperspectral measurements for this analysis.
second group of spectral variables relates the characteristics
and pigment status (primarily chlorophyll) of leaves among the
difference species and its spectral variables consist of C-1D,
A-ID, B-1D, R550, A-WP, SIPI, NPQI, LCI, B-WP, SR, J-1D,
NPCI C-WP, R680, H-1D, and F-1D.
4.2 Species Recognition
To train and test the three-layer ANN structure for classifying
the 11 species (including five oak species), the input of 30
selected spectral variables was first normalized to the range of
[0, 1]. The output layer had 11 nodes corresponding 11 species
(or 5 nodes for 5 oak species). To find a better ANN structure,
various combinations of learning rate (r|), momentum
coefficient (a) and number of nodes in a hidden layer (hi) were
tested using the first training/test data set (Table 1). In
considering relatively small variation of OAA values with all
testing nodes (hi: 10 - 40) and convenience to design the ANN
networks, for identifying the 11 species, all ANNs use hi = 25
or 30, r| = 0.8 or 0.7, and a = 0.1 Or 0.2 while for identifying 5
oak species, all ANNs use hi = 16 or 20, r\ = 0.8 or 0.7, a =
0.1 or 0.2.
For classifying both 11 species and a subset of 5 oak species,
the first row of Table 3 shows classification results calculated
from three sets of test samples by ANN. From the table, we can
see that the classification accuracies (OAA) of averaging three